This repository curates resources on Graph Retrieval-Augmented Generation (GraphRAG), a paradigm enhancing LLMs by integrating knowledge graphs. It targets researchers and developers seeking to leverage structured knowledge for more accurate and context-aware LLM applications, addressing limitations of traditional RAG.

How It Works

GraphRAG enhances traditional RAG by using graph-structured knowledge. This approach explicitly captures entity relationships and domain hierarchies, enabling graph-aware retrieval mechanisms for multi-hop reasoning and context-preserving knowledge acquisition. Structure-guided search algorithms ensure efficient retrieval across large corpora, offering fine-grained, domain-specific information compared to chunk-based RAG.

Quick Start & Requirements

This repository is a curated list of research papers, benchmarks, and open-source projects. It does not provide a direct installation or execution method. Users are directed to individual project repositories for setup and requirements.

Highlighted Details

• Comprehensive categorization of GraphRAG research into Knowledge Organization, Retrieval, and Integration.
• Extensive lists of relevant papers, benchmarks (e.g., DIGIMON, CWQ, MetaQA), and open-source projects (e.g., Microsoft-GraphRAG, Agentic-RAG).
• Includes a survey paper citation for the curated content.

Maintenance & Community

The list is continuously updated. Contact information for corresponding authors is provided for contributions and comments. Links to related "Awesome" repositories for LLMs in graph tasks are also included.

Licensing & Compatibility

The repository itself is likely under a permissive license (e.g., MIT, Apache 2.0, given its nature as a curated list), but users must refer to the licenses of individual linked projects for their specific terms and compatibility.

Limitations & Caveats

As a curated list, this repository does not offer direct functionality or code. Users must navigate to external links for actual implementations, which may have varying levels of maturity, documentation, and support. The "2025" dates for some papers suggest ongoing research.